<![CDATA[This study aimed to (1) design and validate an artificial intelligence-assisted Research Project-Based Inquiry (RPBI) learning model and (2) examine its effectiveness in improving the scientific reasoning of academically underprepared students. A quantitative approach was employed using a non-equivalent control group quasi-experimental design involving 60 students assigned proportionally to experimental and control groups. The intervention was implemented over eight instructional sessions in a foundational science course. The AI component functioned to generate adaptive scaffolding questions, provide hypothesis-checking prompts, and deliver rubric-based formative feedback during project development. Scientific reasoning was measured using a test constructed around five indicators: Variable Identification, Hypothesis Formulation, Data Interpretation, Conclusion Drawing, and Procedure Evaluation. Rasch analysis indicated that the instrument demonstrated satisfactory measurement quality (Person Reliability = 0.82; Item Reliability = 0.91), with appropriate item-person alignment. The validity of the learning model itself was established through expert validation using a structured rubric, pilot implementation feedback, and monitoring of instructional fidelity. Independent sample t-tests revealed statistically significant differences between the experimental and control groups across all indicators (p < 0.001). Exploratory relationship analysis further indicated that Hypothesis Formulation showed the strongest association with other reasoning components, suggesting a hierarchical structure within students’ scientific reasoning processes. These findings indicate that the AI-assisted RPBI model is pedagogically feasible and effective in enhancing scientific reasoning, particularly among students with initial academic gaps.]]>
Copyrights © 2026
